Generator Relaying Operating Due to Leading PF

[trosepe]

I have been asked to investigate two cases where a 3MW generator circuit breaker has tripped on loss of excitation and reverse power when the connected UPS is placed in bypass. The UPS load is 2MW and the loads are PDU's for a server farm. When the system is operating normally the PF is .95 lag. When in bypass the PF is .85 lead. There are no PF correction capacitors on the system and only the one generator. The transfer switch is open transition.

My theory is that the UPS controls the system PF by adjusting the firing angle of the SCR's. When the system is placed in bypass, the PDU's are connected directly to the supply. The leading PF is from the filter capacitance on the front end of the switching power supplies on the blade servers. When the UPS is connected to the generator and then placed in bypass; the generator voltage regulator has a problem with the leading PF and underexcites the field.

I don't have much expertise in generator control to know if I am close or should be looking at something else. Does anyone have experience with how generator regulators respond to leading PF and if the loss of excitation and reverse power(set at 35KW) relay operation makes sense here.

Also any suggestions on a fix? One suggestion has been to connect a reactive load bank to the generator that is connected anytime the UPS is placed in bypass while connected to the generator. The reactance level would be selected to bring the PF back to .95 lag when on generator.

 

[slavag]

Hi.
It ssems as not correct setting of the reverse power and loss of excitation protection. Please check operation angels of those functions.
Did you check those function in the on-load tests?
Best Regards.
Slava

 

[trosepe]

slavag,
The generator vendor maintains the settings and operation are correct. My first recommendation is to have third party relay testing agency test the very things you mentioned. The relay is an SEL 300G. I have checked the CT and PT connections / polarities and all seem correct. Thank you for your post.

 

[slavag]

Hi Trosepe.
I suggest, that your CT/PT connection is OK, let say 90%.
Reason is right operation w/o UPS and of course start of generator.
But..operated two functions with angle settings.
and.. reverse power build on the reverse active power operation, loss of excitation on the "reverse" reactive power.
Reverse power is very sensitive, loss of excitation need a "big" current.
Reverse power operated in the 2-nd and 3-rd Q, loss of excitation in the 3-rd and 4-th Q.

From my expirience, reverse power, in case of "big" reactive current operated only in case of wrong setting, angle setting. Same with loss of excitation.

Please send us to Forum, settings of both functions. SEL300G, I don't know, but here, we have a lot of specialist of this relay.

Best Regards.
Slava

 

[Handlin]

Interesting problem. I do not understand the reverse power trip unless you have multiple generators. Generators react poorly to leading power factor. This is interpreted by the generator controls as the generator being over excited, and lowers the excitation voltage until control of the generator is lost. Cummins has a great white paper describing this sequence of events.

I have seen the problem you describe occur when the UPS is on inverter and the transfer switch transfers between two viable sources - The open transition switch operates too fast. You should always use a delayed or closed transition transfer switch ahead of a UPS.

The UPS on inverter will have a lagging PF at the rectifier input. The load determines the output power factor. Today's dual corded servers have two power supplies. Each supply is power factor corrected - for 100% load. I have seen several data center server loads at leading PF. I have not seen any server load feed by PDU transformers with that low of a leading PF. Lowest server load PF I have seen is .96 leading. A .85 leading PF can not be supported by standby generator

 

[alehman]

I've had problems with diesel gensets tripping with "loss of excitation" when serving leading power factor load. Usually what has really happened is the voltage regulator drives the field current down to zero while trying to maintain output voltage, which causes the reported "loss of excitation" condition.

At this point in effect the voltage regulator has lost control and the unit should be tripped. Your power factor may be outside of the capability curve of the machine, in which case the only way to fix it is to improve the power factor of the load. Can you add some motor loads (e.g. chillers) to the generator?

Alan
----
"It’s always fun to do the impossible." - Walt Disney

 

[waross]

If you add series reactors, investigate the possibility of installing them in the bypass circuit so they will only be in the circuit when they are needed.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[trosepe]

As always great feedback and insights. Thank you all so much. I will post the results of the investigation and the resolution.

 

[ScottyUK]

Light loads combined with heavy reactive loads can cause mis-operation of reverse power relays with very sensitive settings. Phase errors exist in the instrument transformers which in the right (wrong?) combination can result in 'phantom' active power on the secondary side even though the primary side is purely reactive. With enough reactive load and a sufficiently sensitive relay this can cause the relay to mis-operate. Normally this affects steam turbines where the very low motoring load presented by the turbine in vacuum is masked by the combined effect of the high reactive load and phase shifts in the instrument transformers resulting in the relay not operating and the machine continuing to run as a motor. On a diesel set I'd not expect to see this because the reverse power relay can be set to a relatively high setting.



----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[slavag]

Yep, Scotty, you are absolutly right.
for DG ,about 4%P, for big steam 0.5%P, for avois this problem we use meas CT core for the reverse power protection, for 3MW steam about 1-1.5% P.

Best Regards.
Slava

 

[rbulsara]

What supports the cooling load of the server farm, when the UPS is placed on generator?

 

[trosepe]

The HVAC load has a separate generator set. The'building' power and 'data' power systems are kept separate with no means of inter-connection. The building power system operates at .82 lag and has not experienced the RP or LOE relay operations. The designers of the plant felt this was the best way to protect the integrity of the data power system.

 

[trosepe]

waross,

Could you explain why you suggest series reactors as opposed to parallel reactors.

 

[rbulsara]

The'building' power and 'data' power systems are kept separate with no means of inter-connection.

There lies the problem! Feeding just the computers without air conditioning is useless. A bad design to begin with and if you had the related a/c loads on the same generator bus you would not have this issue. Net PF will almost certainly be lagging.

 

[davidbeach]

As rbulsara suggests, it is worthless to have generation backup for the "data" power if you don't have backup for the "building" power. Two, unconnected, generators actually reduces your reliability since you have to have both every time and there are more parts to fail with two than there are with one.

I did a full blown commissioning test of a data center system once, 300kW worth of 1.5kW space heaters as load in the room during commissioning. For one of the tests we tripped the normal power source and blocked start of the generator. That test was aborted long before the UPS batteries were discharged because the temperature in the room was starting to run away. You don't have long.

 

[trosepe]

I agree with you all. I do much design work for telecommunication Central Office and data centers. The data and communication systems are on segregated distribution and transfer systems but combine at the utility and generator busses. Segregating the sources may have seemed like a good idea to the designers but I am sure they will rethink that approach next time. Thanks for all the good input, or is it ouput?

 

[ScottyUK]

David,

We did it for real at my old place - ended up with the doors to the blast-resistant control room wide open for ventilation to avoid tripping the whole plant. The battery would have held up for a few hours but we were in big trouble after about 15 minutes. The whole UPS / generator / source transfer system was replaced a year later.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[davidbeach]

I think it is missed all too often. Computers on UPS and room HVAC left unaccounted for. In that case don't bother with a generator to back up the UPS, just scramble to get everything shut down before the heat drives you out.

 

[rbulsara]

and don't hire the same designers for critical facilities!

 

[alehman]

Another item that's commonly missed is the power supply for HVAC controls and dampers. It doesn't do any good to have the HVAC running if there's no power to keep the dampers open.

Alan
----
"It’s always fun to do the impossible." - Walt Disney